Sensor-based fall risk assessment--an expert 'to go'.

BACKGROUND: Falls are a predominant problem in our aging society, often leading to severe somatic and psychological consequences, and having an incidence of about 30% in the group of persons aged 65 years or above. In order to identify persons at risk, many assessment tools and tests have been developed, but most of these have to be conducted in a supervised setting and are dependent on an expert rater. OBJECTIVES: The overall aim of our research work is to develop an objective and unobtrusive method to determine individual fall risk based on the use of motion sensor data. The aims of our work for this paper are to derive a fall risk model based on sensor data that may potentially be measured during typical activities of daily life (aim #1), and to evaluate the resulting model with data from a one-year follow-up study (aim #2). METHODS: A sample of n = 119 geriatric inpatients wore an accelerometer on the waist during a Timed 'Up & Go' test and a 20 m walk. Fifty patients were included in a one-year follow-up study, assessing fall events and scoring average physical activity at home in telephone interviews. The sensor data were processed to extract gait and dynamic balance parameters, from which four fall risk models--two classification trees and two logistic regression models--were computed: models CT#1 and SL#1 using accelerometer data only, models CT#2 and SL#2 including the physical activity score. The risk models were evaluated in a ten-times tenfold cross-validation procedure, calculating sensitivity (SENS), specificity (SPEC), positive and negative predictive values (PPV, NPV), classification accuracy, area under the curve (AUC) and the Brier score. RESULTS: Both classification trees show a fair to good performance (models CT#1/CT#2): SENS 74%/58%, SPEC 96%/82%, PPV 92%/ 74%, NPV 77%/82%, accuracy 80%/78%, AUC 0.83/0.87 and Brier scores 0.14/0.14. The logistic regression models (SL#1/SL#2) perform worse: SENS 42%/58%, SPEC 82%/ 78%, PPV 62%/65%, NPV 67%/72%, accuracy 65%/70%, AUC 0.65/0.72 and Brier scores 0.23/0.21. CONCLUSIONS: Our results suggest that accelerometer data may be used to predict falls in an unsupervised setting. Furthermore, the parameters used for prediction are measurable with an unobtrusive sensor device during normal activities of daily living. These promising results have to be validated in a larger, long-term prospective trial.

Predicting in-patient falls in a geriatric clinic: a clinical study combining assessment data and simple sensory gait measurements.

BACKGROUND: Falls are among the predominant causes for morbidity and mortality in elderly persons and occur most often in geriatric clinics. Despite several studies that have identified parameters associated with elderly patients' fall risk, prediction models -- e.g., based on geriatric assessment data -- are currently not used on a regular basis. Furthermore, technical aids to objectively assess mobility-associated parameters are currently not used. OBJECTIVES: To assess group differences in clinical as well as common geriatric assessment data and sensory gait measurements between fallers and non-fallers in a geriatric sample, and to derive and compare two prediction models based on assessment data alone (model #1) and added sensory measurement data (model #2). METHODS: For a sample of n=110 geriatric in-patients (81 women, 29 men) the following fall risk-associated assessments were performed: Timed 'Up & Go' (TUG) test, STRATIFY score and Barthel index. During the TUG test the subjects wore a triaxial accelerometer, and sensory gait parameters were extracted from the data recorded. Group differences between fallers (n=26) and non-fallers (n=84) were compared using Student's t-test. Two classification tree prediction models were computed and compared. RESULTS: Significant differences between the two groups were found for the following parameters: time to complete the TUG test, transfer item (Barthel), recent falls (STRATIFY), pelvic sway while walking and step length. Prediction model #1 (using common assessment data only) showed a sensitivity of 38.5% and a specificity of 97.6%, prediction model #2 (assessment data plus sensory gait parameters) performed with 57.7% and 100%, respectively. DISCUSSION AND CONCLUSION: Significant differences between fallers and non-fallers among geriatric in-patients can be detected for several assessment subscores as well as parameters recorded by simple accelerometric measurements during a common mobility test. Existing geriatric assessment data may be used for falls prediction on a regular basis. Adding sensory data improves the specificity of our test markedly.

[Reliability of basic geriatric assessment]. / Untersuchung zur Reliabilität des Geriatrischen Basisassessments.

This study investigates the Geriatric Basis Assessment (GBA) in terms of its reliability. Data from 1037 patients were collected. The reliability was estimated relating to the lambda 2 coefficient. It is necessary to define the items in different categories: the first variable means valuation 1 of each item and not 2, 3, 4; the second variable means valuation 1, 2 against 3, 4; the third variable means the valuation 1, 2, 3 and not 4. The table shows only little difference concerning the lambda 2 coefficients. In conclusion, 80% of the variability of the GBA items can be explained by differences in the patients themselves, while 20% is due to the inaccurate assessment system. For 343 patients, data for both Barthel index and GBA were available. As presumed, correlations between Barthel and connected GBA items were observed. However, the correlations were too weak to predict the Barthel scores from the corresponding GBA item accurately enough. The Barthel index appears to include similar, but not exactly the same aspects as the GBA. The reliability of the Barthel index (lambda 2 = 0.89 for the first variable) is slightly higher compared to the GBA but it is not suitable as a criterion of validity. Both the validity of the GBA and the Barthel index can not be determined lacking an external measure. As an example, a suitable criterion of validity could be the reintegration into the familiar surroundings preceding the hospital stay. When developing the GBA, it was not assumed that geriatric patients could be correctly diagnosed on the basis of an overall score alone or to allocate them to adequate care using that score as a sole indicator. Crucial for these purposes is the test profile as a whole, including the impairments, disabilities handicaps, and last but not least the diseases of the individual patient. Furthermore, the depiction of the GBA profile at admission and discharge allows one to identify those items, on which therapy has a significant influence and those which remain more or less stable. As presumed, items with minor initial deficits (e.g., motivation, eyesight, hearing, depression, capability of verbal expression, situative adaptability, understanding) showed only small differences between admission and discharge. On the other hand, items strongly influenced by geriatric treatment were, e.g., mobility (walking, transfer), functions of internal medicine, and domestic care. Prognostically significant are those items which are crucial for reintegration and describe a deficiency but cannot be altered reliably. Such items are the person, to whom the patient relates most closely, situative adaptability, motivation, orientation, capability of verbal expression, and possibly depression. All of these parameters are more difficult to influence than the activities of daily living assessed by the Barthel index. Further investigations should clarify whether the GBA can be a reliable tool for allocating a patient to adequate care. However, the requirement for such a criterion of validity is that this allocation is truly optimal for the patient.